TECHNICAL NOTE

Implementing the Requirements of the European FloodDirective: the Case of Ungauged and Poorly GaugedWatersheds

S. Yannopoulos1 & E. Eleftheriadou2 & S. Mpouri1 &

Iο. Giannopoulou1

Received: 6 February 2015 /Accepted: 22 July 2015 /Published online: 2 August 2015# Springer International Publishing Switzerland 2015

Abstract In 2007, the European Floods Directive (FD) 2007/60/EC came into force, intro-ducing a framework for the assessment and management of flood risks. According to Article 6of the Directive, Member States shall prepare flood hazard maps and flood risk maps at acatchment level, covering the areas that could be flooded under different probability scenarios.The former maps include crucial information towards flood management, such as flood extent- water levels - flow velocity, that will form the base of the flood risk management plans.Based on these data, Member States shall set their objectives and prepare measures concerningtheir flood management plans. Hence, it is obvious that effective flood management requiresreliable methods for the estimation of the flooded areas, such as precise estimation of the peakdischarge. This is a hard task for hydrological engineering, taking into account that in themajority of small catchments, especially in Greece, there is a substantial lack of hydrologicaldata. The present paper presents the effect of the uncertainty, derived from lack of data, on theestimation of the peak discharge based upon which the flooded areas will be identified.Towards this direction, a methodology of peak discharge estimation is presented in anungauged small catchment followed by hydraulic calculations using HEC-RAS and HEC-GeoRAS software packages. The results demonstrate the great variability of the estimatedflood extent and the effect of it on the decision-making of the proposed flood measures.

Keywords Ungauged watershed . Directive 2007/60/EC . Estimation peak discharge . Floodmaps

Environ. Process. (2015) 2 (Suppl 1):S191–S207DOI 10.1007/s40710-015-0094-2

* S. Yannopoulosgiann@vergina.eng.auth.gr

1 Laboratory of Hydraulics Works and Environmental Management, School of Rural and SurveyingEngineering, Faculty of Engineering, Aristotle University of Thessaloniki, GR-54124 Thessaloniki,Greece

2 Directorate of Environment and Spatial Planning, Region of Central Macedonia, Macedonia, Greece

1 Introduction

It is considered that due to climate change, flooding events have become more frequent in therecent years. In order to face this fact, the European Union issued the Floods Directive (FD)(2007/60/EC) (Council of the European Communities 2007), as a tool to assist Member Statesto manage the floods effectively. Each member should comply with the requirements as set inthe Directive and achieve certain goals within a strict timeline.

The Floods Directive 2007/60 entered into force in November 2007 and since then itprovides a complementary part of the Water Framework Directive (WFD) 2000/60/EC,concerning flood issues. Its primary goal is to enact a framework for the assessment and themanagement of flood risks targeting at the reduction of the negative effects on human health,the environment, the cultural heritage and economic activities that are connected with floods(Article 1).

Effective flood management is achieved through the detailed knowledge of the hazards andrisks that exist in a particular area. In order to organize this crucial data, flood maps areindispensable tools to show information about hazards, vulnerabilities and risks in a particulararea (EXCIMAP 2007). Moreover, the geographical identification of areas at different level ofrisk through flood mapping, make the information more transparent and easily accessible tothe public. Flood mapping plays a crucial role on the understanding of flood risks and providesa reliable base upon which each Member State will decide on their actions to avoid, mitigate,transfer, share, or accept the risks. According to Tsakiris et al. (2009), a good starting point forthe preparation of flood risk maps is the confirmation of the safety standards by differentauthorities (state, regional or local), and also considering different safety standards for differentcauses of flooding, facilitates flood risk management, where the term “safety standard” meansthe exceedance frequency of water level. The FD only mentions (Article 6) three probabilitiesof flooding (high, medium and low) and it leaves the choice of the flood exceedance frequencyof water level (safety standard) to Member States. In our view, an in depth analysis concerningthe flood risk issues is required, regarding safety standards and approaches, and mainly,whether these should be set out at the EU level or at a national, regional and river basin level.Concerning the identification of these measures, the FD renders responsible each MemberState by stating “detailed objectives for protection against floods, measures best suited toachieve the objectives and deadlines will not be defined at EU level” (COM 2006).

A reality that often has to be faced by Member States, especially when dealing with smallcatchments, is the lack of hydrologic data that is essential for the subsequent flood mapping.Especially in Greece, where the majority of the country’s territory comprises small ungaugedand poorly gauged watersheds, flood estimation and design of proposed flood preventionschemes is a hard task. Moreover, many methods often used (such as SCS-CN), are based onfield data from few experimental catchments and are not updated nor validated, affecting thecost and safety of flood protection works (Efstratiadis et al. 2013).

In the present paper, the case of a small ungauged watershed in North Greece (Tourlawatershed) is investigated, in relation to the estimation of the peak flow using a set of variousmethodologies. The results of the hydrologic calculations of all the methods are then intro-duced in the hydraulic model HEC-RAS, in order to demonstrate the effect of the uncertaintycharacterising ungauged watersheds on flood mapping. Flood hazard maps should include datafor three scenarios of flooding (low-medium-high probability), but in the current study only thehigh probability is demonstrated, as the Greek technical specifications were applied, where thereturn period when designing flood prevention works should be set to 50 years.

S192 S. Yannopoulos et al.

2 Legislation

The proposition for the preparation of a European Directive for the management of floods wasfirst mentioned in the Floods Action Programme prepared by the European Commission. Thefirst draft of the FD was released in January 2006 after a public consultation. The negotiationsof theMember States concluded unanimously in the final text of the Directive on 27 June 2006,that was later adopted as Common Position of the Council on 18 October 2006. Directive2007/60 entered into force in November 2007 and since then it provides a complementary partof the WFD concerning flood issues.

The most important elements of Directive 2007/60 are:

– Framework on the assessment and management of flood risks.– Preliminary flood risk assessment based on available information and experience from

past floods concluding in predictions of future floods and identification areas of potentialsignificant flood risk.

– Flood mapping: flood hazard maps and flood risk maps according to different scenarios(low – medium – high probability).

– Establishment of Flood Risk Management Plans, where the Member States shall establishappropriate objectives focusing on the reduction of adverse consequences of flooding forhuman health, the environment, cultural heritage and economic activities.

– The encouragement for the establishment of an international flood risk management planor a set of plans coordinated at an international catchment level.

– Member States shall coordinate the implementation of the FD Directive with Directive2000/60. Both river basin and flood risk management plans are fundamental for theintegrated basin management and thus every possibility of synergy between them shouldbe exploited.

– The active involvement of all interested parties is pursued and coordinated with article 10of Directive 2000/60.

In 2010, the FD was introduced in the Greek legislation with a Common MinisterialDecision (2010) and in December 2012 the report of the preliminary flood risk assessmentwas issued, according to Article 4 of the FD. The report included maps of all the river basindistricts, data collection of historic floods and identification of those with significant adverseimpact, taking into account any human losses, the compensation provided and the flooded area(Special Secretariat for Water 2012).

The main types of flood maps introduced by the FD are flood hazard and flood risk maps,where the former contain information on the probability and magnitude of flooding and thelatter incorporates the consequences of flooding. Flood extent maps are the most widespreadflood maps, followed by flood depth maps, sometimes velocity and rarely propagation (deMoel 2009). Flood hazard maps include crucial information towards flood management, suchas flood extent - water levels - flow velocity, that will form the base of the flood riskmanagement plans. Based on these data, Member States shall set their objectives and preparemeasures concerning their flood management plans.

The European Commission, acknowledging the importance of the preparation of reliableflood maps, issued in 2007 the “Handbook on good practices for flood mapping in Europe”(EXCIMAP 2007). According to this document, flood maps form the basis for the manage-ment of flood risks and they form a prerequisite for achieving effective and efficient flood

Implementing the Requirements of the European Flood Directive S193

management. The status of flood mapping in Europe is described in a document prepared forthe 2015 EU Water Conference (European Commission 2014), reporting on the progress ofMember States on flood mapping by December 2014, where information is missing fromBulgaria, Greece, Malta and Portugal.

The subsequent step of flood risk mapping varies from the flood hazard mapping, as itincorporates more indicators for the assessment of flood consequences compared to the fewindicators used during flood hazard estimation (de Moel 2009). Moreover, flood risk mappingis highly affected by the questions that have to be addressed. According to FD: “Flood risk isthe combination of the probability of a flood event and of the potential adverse consequencesto human health, the environment and economic activity associated with a flood event”. Riskdoes not remain constant in time, as the vulnerability parameters (such as population, assetsand economic activity) may change rapidly, and it is often necessary to predict changes in riskin the future, to make better decisions (Handbook on Good Practices 2007). Flood risk mapsare qualitative maps, which show the indicative number of inhabitants potentially affected, thetype of economic activity affected such as installations (industrial activities) of Annex I of theDirective 96/61/EC and the protected areas of Annex IVof the WFD. The level of detail andthe type of information depicted on the maps can be decided upon the special features of thearea under study and any particular interest of the Member State (e.g., distribution of particularvulnerable groups or affected installations causing pollution).

Flood Risk Maps play an important role in the decision-making process, as it identifies theareas where the greatest risk exist and sets the priorities concerning the planning of the floodprotection measures. Moreover, such maps can support the decision-makers with the selectionof the best combination of measures, the designation of flood warning systems and even withthe management emergency situations and crisis management (Handbook on Good Practices2007). Hence, the identification of the flooded areas (flood extent, water depth), during theprecedent process of flood hazard mapping, forms the basis for the subsequent requirements ofthe FD, as it affects directly the preparation of flood risk maps and consequently the FloodRisk Management Plans.

3 Methodology

3.1 Introduction

In the current study, a set of various methodologies of hydrograph estimation arepresented and compared in order to investigate the variations of the peak dischargeand examine which factors affect and at what level the produced results. A similarstudy by Yannopoulos et al. (2012), showed great variations of the produced designhydrographs, especially on their geometric characteristics, where four different statis-tical models were used for the Intensity-Duration-Frequency (I-D-F) relationship. Inthe present study, the I-D-F equation was derived at an adjacent station (Papamichailand Georgiou 1999). In order to transfer information between watersheds or evenbetween regions, climatic, geologic and geomorphologic homogeneity between thewatersheds or the regions are needed (Brass 1990). According to Pilgrim and Cordery(1993), “this highlights the danger of transposing procedures and design informationfrom one country or region to another and reinforces the need for the derivation ofprocedures and parameters from observed data in each region”. To the extent that we

S194 S. Yannopoulos et al.

keep in mind, there are not such conditions in the wider region of the study area norsimultaneous rainfall-runoff measurements.

In another study by Yannopoulos et al. (2014), the effects of the Synthetic UnitHydrograph (SUH) selected (SCS and Sierra Nevada SUHs) and the storm distributionmethod on the hydrographs produced, demonstrate substantial differences on peak dis-charge and time of peak. Thus, the selection of the most appropriate methodology forms acrucial step in flood studies, as the proposed flood protection works are designed based onthe predicted water level that corresponds to the estimated peak discharge, and conse-quently, any uncertainty in the estimation procedure affects directly the cost effectivedesign of the flood protection works.

The standing technical specifications in Greece (Presidential Decree 696/1974, Arti-cles 177–194) concern design of hydraulic works and mainly irrigation projects (Article177§1), but they can be applied also in hydrological studies after the necessary amend-ments are made (Article 194). As far as it concerns the issue of design flood flows,certain methods can be used depending on the nature and the importance of the projectsunder study. According to the legal instructions, the following calculation methods canbe applied:

Empirical methods of estimating maximum flood flow.Statistical methods based on the statistical processing of flow and water stage measure-ments in order to extend their application in cases of higher return periods.The Rational Method.Time - Area method.Hydro-meteorological methods such as the unit hydrograph method or use of precipita-tion data from adjacent watersheds under similar hydrologic conditions.

Eight cases were investigated in the current study, in order to estimate the effect ofeach method on the outcome concerning peak discharge, time to peak and total floodvolume. As shown in Table 1, each case is characterised by a specific storm duration, astorm distribution (Alternating Block Method and SCS method for type I and IA) and aSUH (SCS and Sierra Nevada).

Table 1 The eight cases investigated

Case Storm Duration (hrs) Storm Distribution S.U.H.

6 12 24 Alternating Block Method SCS SCS Sierra Nevada

Α √ √ √ √ √Β √ √ √ √ √C √ √ √D √ √ √E √ √ (I) √F √ √ (IA) √G √ √ (I) √H √ √ (IA) √

Implementing the Requirements of the European Flood Directive S195

3.2 Design Flood Hydrographs

According to Westphal (2001) “when watersheds are large, that is, when they are comprised oftwo or more smaller watersheds whose streamflow at the confluence with common collectorchannel can be expected to be displaced in time, where storage influences the time distributionof flow in a stream, or where storage is a part of the design problem, peak flow methods areinappropriate for hydrologic design. In these instances, it is necessary to estimate the entireflow hydrograph.” In the current study the produced design flood hydrographs from differentmethods are investigated, in order to compare not only peak flows, but also all geometriccharacteristics that are affected by the various methodologies.

3.2.1 Estimation of the Design Hyetograph

Return Period The choice of the return period of the design flow is of paramount importanceand it depends on various factors such as the size of the watershed, the accepted failure risk andthe type of the designed structure. In practice, for small watersheds short return periods areselected whereas long ones are selected in the case of structures such as levees (between 50and 100 years). As a rule of thumb, in areas where no direct risk on properties exist, a returnperiod of 50 years is considered safe, where in areas with human population the return periodshould be increased at 100 or 200 years.

It is obvious that increasing return period represents an increase on project costs, buton the other hand reduces the potential damage from any construction failure. Theoptimal return period can be derived by optimizing the sum of the construction costsand the cost of the remediation works. In practice, it is almost impossible to quantify theamount of damages, so the choice of return period in Greece follows in general thefollowing principles (Ministry of Environment 2002a): i) for storm sewers in residentialareas: T=2 to 15 years (most frequent value is 5 years); ii) for storm sewers incommercial areas and for central sewer collectors: T=10 to 15 years; iii) for floodcontrol projects and settlement on watercourses: T=50 years and more.

In cases of constructions near watercourses (bridges, culverts), it is necessary to estimate thedesign flood hydrograph, which is connected directly with the design storm. In general, thereturn period of a design storm does not coincide with the return period of the correspondingflood design (e.g., flood peak), mainly due to the impact of hydrological losses (retention,infiltration, etc.) that may differ during each storm. According to the Ministry of Environment(2002b), it is preferable to settle the return period of the design storm from the return period offlood, since it is more usually met that rainfall data covers long time periods and present morecomplete records, even in cases where water flow measurements do exist.

According to Larson and Reich (1973), this differentiation between the return period of thedesign flood and the design storm is not appropriate, since on average the two periodscoincide. Moreover, Koutsoyiannis (2004) argues that such a distinction should not be takeninto account. In the present study, the return period of the design storm was calculated from theone of the design flood according to Sutcliffe (1978). Thus, a return period of 50 years wasselected based on the nature of the project under study –river bounding- that corresponds to areturn period of 81 years design storm.

Storm Duration Defining storm duration is a crucial step of the peak discharge estimationprocess. As a general rule, the duration of the design storm should be at least equal to the time

S196 S. Yannopoulos et al.

of concentration in order to be able to calculate peak discharge (Westphal 2001). Usually, theduration is taken as multiple of 3 or 6 h. In order to select the most appropriate duration,several trials should be made in order to choose the one that leads to the worst outcome.However, as shown by the analysis hydrographs, increased flood peak appears usually forshort durations (short duration corresponds to a large volume). In general, a few tests should bealways run in order to determine the duration of the design rainfall (Tsakiris 1995).

According to Tsakiris (1995), for projects as drainage networks and storage projects (e.g.,dams) in small watersheds, the hydraulic calculations should be realized for rainfall durationsless than 3 h. Likewise, for large storage projects or for major flood control projects longerrainfall durations should be applied (e.g., 6, 12, 18 and 24 h). In the latter case, specialattention must be drawn on the temporal distribution of rainfall considered.

According to the Greek Ministry of Environment (2002b), rainfalls lasting 24 h is areasonable choice for catchments where rainfall duration is longer than the time of concentra-tion. It should be noted that during the process of determining the design storm, the stormduration should be taken as a significant multiple (well over double) of the watershed’s lagtime. The latter view was followed in the present study.

Rainfall Distribution Estimating the design flood while considering constant rain intensitythroughout its duration could easily lead to false conclusions, as rainfalls with duration of afew hours are considered too long to assume that the intensity remains constant. From theliterature it is shown that the temporal distribution of rainfall is important for the resultingflood hydrograph. Indeed, two storms with the same depth but with different distribution forthe same duration produce different flood hydrographs.

According to McCuen (1998): “for many problems in hydrologic design it is necessary toshow the variation of the rainfall volume with time. That is, some hydrologic design problemsrequire the storm input to the design method to be expressed as a hyetograph and not just as atotal volume of the storm”.

Three methods were applied in the current study for the calculation of the stormdistribution; i) the Soil Conservation Service (SCS) method for storm duration of 6 h(SCS 1972), ii) the Soil Conservation Service, for storm type I and IA of 24 h durationand iii) the Alternating Block Method (ABM) for storm durations of 6, 12 and 24 h(Chow et al. 1988). Obviously, the SCS method was developed for conditions met in theUSA, while there is no bibliographic reference that examines which storm types may beapplicable to the Greek conditions. For this reason, and due to lack of data, methods i)and ii) were not evaluated further in the present study.

Rainfall Depth Proper design of water resources projects is highly based on the Intensity– Duration – Frequency relationships that are extracted from historic data of extremerainfalls of different durations, while the return periods used should be equal to theeconomic projects’ life. Estimating the maximum rainfall intensity (i) for various durations(t) and for various return periods (T) is achieved through the frequency analysis of extremevalues. The maximum values of rainfall intensity normally follow a frequency distributionof extreme values, which is not easy to define. Various applications of distributions havebeen observed over time. In the current study four statistical distribution models areapplied; Gumbel, Pearson III, Pearson III using the frequency factor and log-Pearson III.The assessment of rainfall losses and consequently of the effective rain was realized by themethod of the runoff coefficient CN of the SCS method.

Implementing the Requirements of the European Flood Directive S197

Rainfall data is taken by a meteorological station installed in an adjacent catchmentarea, where the I-D-F relationships and the corresponding coefficient of determination,R2, have been derived by Papamichail and Georgiou (1999). Equation (1) is based onGumbel distribution with a high coefficient of determination R2=0.9636, and thus, it isconsidered that it represents quite satisfactorily the frequency of the maximum annualrainfall values observed in the area (as registered at the local meteorological station‘Megali Panagia’).

Gumbel : i ¼ 20; 5069T0;1613

t0;64R2 ¼ 0:9636 ð1Þ

3.2.2 Determination of the Synthetic Unit Hydrograph

In the absence of appropriate measurements of runoff and rainfall, the unit hydrograph can becalculated using the geometric characteristics of the watershed under study. The most popularSUH is the one developed by the Soil Conservation Service (SCS), which was used in thecurrent study and was compared with the Sierra Nevada SUH as described in the Design ofSmall Dams (U.S.B.R. 1987).

3.3 The Study Area

The study area is situated in the southern part of Sithonia Peninsula at Chalkidikiregion in Northern Greece. The land cover is characterized by mountainous and hillyareas, covered with low vegetation and pine forests. The climatological conditions arerepresentative of the typical Mediterranean climate due to the ground relief and thevicinity of the area to the sea. The mean annual precipitation is 477.9 mm, whereasprecipitation records show that high values are observed during the period of 1 day,with the observed maximum value taken place in June 1975 reaching 90.3 mm (Mpouri2008). High values of precipitation are observed in May where the mean monthlyvalue is estimated at 61.4 mm, while the lowest values occur in August (16.2 mm meanvalue) (Fig. 1).

Fig. 1 The study area of Tourla watershed

S198 S. Yannopoulos et al.

3.4 The Hydraulic Model

The hydraulic modelling of the case study (Tourla watershed) was realised using initially theHEC-GeoRAS software developed by U.S. Army Corps of Engineers, in order to import thegeometric characteristics of the basin. The GIS data was then imported in HEC-RAS softwaredeveloped by U.S. Army Corps of Engineers as well, in order to simulate the flood flows. Inthis stage, it was found out that the DEM of the watershed was not sufficient for the detailedmapping of the stream course. In order to tackle this issue, it was decided to use 18 crosssections that were derived from in-situ measurements and were provided to the authors by aprivate company. The boundary conditions were set as known water surface. For the upstreamboundary the critical depth was mathematically calculated (Eq. 2) according to Chanson(1999), whereas the downstream boundary was set at the sea level (0 m). It is noted that theupstream cross section is considered as a triangular shape. The one-dimensional numericalsimulation was run for unsteady flow and mixed flow conditions for all the hydrographs thatwere produced by all the methods.

yc ¼2 Q2

g m2

� �0:2

ð2Þ

4 Results

4.1 Design Flood Hydrograph for Rainfall Distribution of ABM

The ABM provides a simple way to distribute the rainfall by placing the peak of the hyetographat the middle of the storm duration. Initially, the rainfall intensities were calculated using fourdifferent I-D-F curves for design flood return period of 50 years, that corresponds to design rainreturn period of 81 years (Sutcliffe 1978), and for rain durations of 6, 12 and 24 h. Based on theresults, the corresponding depths of the total effective rain were estimated from the equation h=i.t (where i: intensity and t: storm duration). The resulting hyetographs are shown in Fig. 2.

The unit hydrograph is determined on the basis of the peak discharge, Qp, the time topeak, tp, and the dimensionless parameters of time and direct runoff (ta, Qa). Two caseswere investigated and compared, the SUH of the SCS method and the SUH of SierraNevada. Figure 3 demonstrates that the SUH of the SCS method produces greater peakvalues in a shorter time compared with the corresponding values derived from the SierraNevada U.H. Also, the analysis of the above SUHs shows a higher peak flow for thedesign storm of 0.50 h. In particular, for the rain of 0.50 h, the peak flow deviationderiving from the two SUHs is 23 %, while for the time to peak is around 120 %. Thediscrepancies observed in the results are due to the diversity of the assumptions combinedwith the dimensionless unit hydrographs that were used in each SUH method.

The Design Flood Hydrographs derived using the ABM method and for the designstorms of 6, 12 and 24 h and for 50 years return period were calculated based on theprinciples of proportionality and superposition and are depicted in Figs. 5, 6 and 7.

4.2 Design Flood Hydrograph for Rainfall Distribution SCS

The SCS method has developed four storm types (I, IA, II, III) for storm duration of 24 h andone type for storms of 6 h duration. Concerning Greece, there is lack of a systematic study that

Implementing the Requirements of the European Flood Directive S199

would adjust the storm types to the local climatic and environmental conditions. The selectionof the storm types applied in the current study is based on Daniil et al. (2005), where types Iand IAwere used corresponding to wet winters and dry summers, while the produced designhyetographs are shown in Fig. 4. The produced Design Flood Hydrographs are shown inFigs. 5, 6 and 7.

4.3 Comparison of Results

4.3.1 Hydrologic Calculations

In the present study, the variation of the peak flow values of a small watershed in NorthernGreece is investigated, where there is lack of data as in the majority of similar cases. For thispurpose, empirical equations and methods were used as described in the Greek specificationsalong with more theoretical methods, such as the SCS methods and the Sierra Nevada SUHs.

Fig. 2 Synthetic hyetographs using the ABM for storm duration 6 h for four I-D-F curves

Fig. 3 SUH of SCS method and Sierra Nevada for storm duration 0.5 h

S200 S. Yannopoulos et al.

The resulted design hydrographs from the hydrologic calculations prove that there is greatuncertainty of the computed peak discharge, steaming from the selected method applied. Highvariations of peak values and time of peak are observed for all the cases investigated (differentstorm duration, storm distribution and SUH method).

By comparing the results of the eight cases as presented in Table 1, it can be concluded that:

a. The type of the SUH selected affects greatly the peak flow and the corresponding time topeak, while the flood volume is only slightly affected for all the storm durations that wereapplied.

b. The resulting flood hydrographs for storms lasting 24 h (Fig. 7 – Cases E-F and G-H),which are calculated using the same SUH method but different type of storm (I or IA)produce peak flows with differences ranging to 30 %, much less variation for the time topeak and almost none for the flood volume.

c. It is concluded that the SUH used in each case affects significantly the geometriccharacteristics (peak flow variation around 40 % and time of peak 17 %) of the floodhydrographs, due to the assumptions introduced in the methodologies of the SCS and the

Fig. 4 Synthetic hyetographs of the SCS distribution for storm duration of 6 h and four I-D-F curves

Fig. 5 Design hydrographs for storm duration of 6 h for Cases A, B, C and D

Implementing the Requirements of the European Flood Directive S201

Sierra Nevada SUH (Fig. 5), while the method of the storm distribution does not cause thatsignificant changes in the hydrographs.

d. The hydrographs of Cases A and E give the highest peak flow values, while variations ofthe time of peak are observed for all cases (Fig. 7). Comparing the distribution of stormtype I and IA, it is concluded that the peak value of type I rainfall occurs between 11 and12 h, while for type IA during 9–10 h. The results show that the selection of the stormdistribution of the design rainfall plays a crucial role in the determination of the designflood hydrograph

e. When comparing the two SUHs applied, it is concluded that the SCS method produceshigher peak values than the Sierra Nevada one (as in Fig. 6).

It is clear from Fig. 5 that the higher values of peak flows are observed in the case of theSCS SUH combined with the ABM for storm distribution (Case A), while lower peak valuesare predicted by the Sierra Nevada SUH combined with the SCS storm distribution (Case D).Deviations of peak flows for Cases A and B amount up to 40 %, while similar findings areattributed to time to peak and total flood volume. As for the time to peak, Case C estimates

Fig. 6 Design Hydrographs for storm duration of 12 h for Cases A and B

Fig. 7 Design hydrographs for storm duration of 24 h for Cases A, B, E, F, G and H

S202 S. Yannopoulos et al.

lower values compared to Cases A, B and D. It should be noted that for all Cases the volumedeviations are negligible.

The impact of each selected methodology is illustrated in Figs. 6 and 7 where thehydrographs of the rain of 12 h and 24 h are illustrated. The design hydrographs whenapplying different SUH methods, vary significantly (Fig. 6) concerning their geometriccharacteristics, the peak flow (maximum deviation around 48 %) and time to peak (maximumdeviation approximately 16 %). As shown in Fig. 7, where the methods of SCS and SierraNevada S.U.H. are applied, the combined effect of changing both storm distribution and theSUH method is higher for the peak flow (reaches up to 70 %).

The results show that the selection of the storm distribution, storm duration and SUHmethod,play a crucial role in the determination of the design flood hydrograph. Moreover, concerning anyuncertainty derived by the use of the I-D-F curves, it would be extremely useful that I-D-F curvesare estimated for each catchment of Greece by a central authority that could function in anindependent status and probably under the supervision of the Ministry of Environment.

4.3.2 Hydraulic Model

At a next stage, the produced hydrographs are inserted in the hydraulic model and afterrunning the model for unsteady-state flow conditions, the variation on the flooded area foreach case investigated is depicted clearly, as shown in Fig. 8 in the 3D plot of the water lever ateach cross section. Figure 9 provides a more detailed view of the flooded areas in two parts ofthe stream, where it is shown that hydrographs of Case A and Case H (the two cases where thehighest and the lowest peak discharge is observed), present high variances of the affectedareas. By comparing the results of these two extreme cases, it is observed that “Flow Area” canvary up to 84.7 %, “Water level” up to 44.4 %, “Top Width” up to 85.9 % and “Velocity”values may differ more than 100 %.

For instance, as shown in the following Fig. 9, top width of the inundated area can varyfrom 45.26 to 11.15 m at cross section 18, while at the residential area further downstream (atcross section 8), variations range from 33.23 to 12.47 m. It is obvious that these variations canbe misleading, when investigating the effects of a flood or when estimating economic figures

Fig. 8 Flooded areas for different cases

Implementing the Requirements of the European Flood Directive S203

(such as cost of flooding during a cost-benefit analysis), either on a rural or an urban area. Inaddition to this, the differences observed in the velocity values, could affect the hydrauliccalculations concerning the estimation of sediment transport in the channel.

Based on the results of the hydraulic model, flood protection works can be proposed at eachcross section, such as cross section widening and levee construction where suitable, as in thedownstream part building structures adjacent to the watercourse confine the selection of floodmeasures. By analyzing the results of the water profile, certain flood measures are proposed foreach part of the stream. Stream-bed settlement and cross section widening are applied in allcross sections in order to increase the flow capacity, whereas widening of the downstreamsections is limited due to the expansion of the residential area. In Fig. 10, the effect of the floodprotection works are shown for Cross section 6, which is situated at the downstream part of thewatershed. As shown, concerning Case A, in order to prevent the adjacent area of flooding,levees of 0.5 m height should be constructed, while the water level of Case H is well below theembankment and hence no flood measure is needed. The above remark proves the necessity ofimplementing reliable methods when estimating the input hydrographs in a model, as theengineer could be misguided and propose unrealistic flood measures.

5 Conclusions

In the present study, the design hydrographs of an ungauged small watershed were estimatedfor different storm durations and distributions, combined with two different SUHs. The resultsshow that peak discharges are highly affected by the methods selected for all the aboveparameters. The return period of 50 years remained constant in all the calculations as it isconsidered a safe choice for river bounding in a non-residential area. The engineers should paygreat attention to the methodology followed and especially to the use of empirical methods,even if they are stated in the Greek specifications, as they should be applied with great cautiondue to the great variability of the resulting peak discharge. A comparison of a number ofselected methods prior to the decision-making phase, could be proven useful, as it wouldprovide the engineers with a variability of results, in order to choose the most appropriate onesin relation to the flood prone area and the designed flood protection measures.

The hydrological analysis demonstrated that the selected methodology of the storm distribu-tion plays a crucial role in the identification of the design hydrograph. For example, type I stormdistribution generates a hyetograph with a peak value at 9–10 h, while for type IA the peak is

Fig. 9 Details of Fig. 8

S204 S. Yannopoulos et al.

observed at 7–8 h and for the ABMmethod at 11–12 h. Hence, it is obvious that the selection ofthe storm distribution method, affects directly the time period of the peak value. In practice, moresimplified methods are preferred (Westphal 2001), such as the ABM, while others based on non-dimensional time distribution are considered as arbitrary (Koutsoyiannis 2004).

The ABMpresents certain advantages compared to the other bibliographicmethods, such as:

& It is based strictly on data of the study area (rainfall intensity curves) rather than ondiagrams or tables of the bibliography.

& It produces a single design hydrograph with no further assumptions.& The produced results seem more reasonable than the ones produced from the other

methods.

However, the basic disadvantage of this particular method is the fact that for each time step,the corresponding rainfall depth has the same return period with the total rainfall depth.

Concerning flood management, the lack of rainfall-runoff data causes many impediments tothe country’s compliance with the FD’s requirements. The preparation of flood hazard mapsthat were completed in March 2014, should have been based on well-structured models, as theresults produced (flood extent, water depths and flow velocity), will form the basis of the floodprotection management plans as described in Chapter IVof the FD. This is a highly importantprocedure as these plans will include measures focusing on the reduction of potential adverseconsequences of flooding for human health, the environment, cultural heritage and economicactivity. Hence, the construction of cost effective flood prevention schemes directly dependson the reliability of these plans.

The preparation of flood hazard maps is not the endpoint of the implementation of the FD.The subsequent step of the preparation of flood risk maps contains many issues to be addressed,such as the definition of risk which may vary greatly among different case studies. Moreover,the interpretation of the above maps will form the base of the Flood Risk Management Plans,where the proposed measures and preparedness against flooding shall be addressed. Hence, thestarting point of all these procedures, i.e., the reliable estimation of hydrologic conditions, formsa crucial point in the successful implementation of the FD across Europe.

Acknowledgments An initial version of this paper has been presented in the 12th International Conference onProtection and Restoration of the Environment, Skiathos Island, Greece, June 29 to July 3, 2014.

Fig. 10 a Cross section 6 before; b after the flood protection works for the two extremes peak discharges (CaseA and Case H)

Implementing the Requirements of the European Flood Directive S205

References

Brass RL (1990) Hydrology: an introduction to hydrologic science. Addison-Wesley Publishimg Company,Reading

Chanson H (1999) The hydraulics of open channel flow: an introduction. basic principles, sediment motion,hydraulic modelling, design of hydraulic structures. Wiley, New York

Chow VT, Maidment DR, Mays LW (1988) Applied hydrology. McGraw-Hill, New YorkCOM (Commission of the European Community) (2006) Proposal for a directive of the European parliament and

the council on the assessment and management of floods. COM(2006)15 final of 18.1.2006Common Ministerial Decision No. 31822/1542/E103 (2010) Assessment and Management of Flood Risks in

compliance with Directive 2007/60/EC of the European Parliament and of the Council of 23 October 2007.Greek Republic, Ministry of Environment, Energy and Climate Change

Council of the European Communities (2007) Directive 2007/60/EC of the European Parliament and of theCouncil of 23 October 2007 on the Assessment and Management of Flood Risks

Daniil EI, Michas SN, Lazaridis LS (2005) Hydrologic modelling for the determination of design discharges inungauged basins. Global Nest 7(3):296–305

de Moel H, van Alphen J, Aerts JCJH (2009) Flood maps in Europe –methods, availability and use. Nat HazardsEarth Syst Sci Discuss 9:289–301

Efstratiadis A, Koussis AD, Koutsoyiannis D, Mamassis N (2013) Flood design recipes vs. reality: canpredictions for ungauged basins be trusted? Nat Hazards Earth Syst Sci Discuss 1:7387–7416

European Commission (2014) Member States’ Examples of Flood Hazard and Flood Risk Maps, available at:http://ec.europa.eu/environment/water/flood_risk/pdf/MS%20 examples.pdf. Accessed 30 Jan 2015

EXCIMAP (European Exchange Circle on Flood Mapping) (2007) Handbook on Good Practices for FloodMapping in Europe, mapping Available at: http://ec.europa.eu/environment/water/flood_risk/flood_atlas/pdf/handbook_goodpractice.pdf. Accessed 29 Jan 2015

Koutsoyiannis D (2004) ‘Research Project: ‘Investigating the stability of slopes and bottom of Filothei streamusing mathematical models and modern environmental techniques’, Scientific Coordinator Stamou A. Vol.Hydrologic Study of floods Athens, National Technical University of Athens

Larson C. L. and B. M. Reich (1973) ‘Relationship of observed rainfall and runoff recurrence intervals. In Floodsand Droughts’, Proc. 2nd Intern. Symp. in Hydrology, Water Resources Publ, Fort Collins, CO

McCuen RH (1998) Hydrologic Analysis and Design. (2nd ed.), Prentice Hall, IncMinistry of Environment, Planning and Public Works D2002a] ‘Guidelines to Design and Construction

Supervision’, Volume 8: Irrigation-Drainage. Hydraulic Works, AthensDin greek]

Ministry of Environment, Planning and Public Works D2002b] ‘Guidelines for Road Works Studies’, Volume 8:Irrigation-Drainage. Hydraulic Works, AthensDin greek]

Mpouri S (2008) Bounding of a river of small watershed without rainfall records using the programs ArcGIS,HEC-GeoRAS and HEC-RAS. Postgraduate thesis, Postgraduates Studies Program ’GeoInformatics’,School of Rural and Surveying Engineering, Faculty of Engineering, Aristotle University of Thessaloniki

Papamichail D and Georgiou P (1999) Statistical analysis of the intensity-duration-frequency relationships ofextreme rainfalls. Proceedings of 12th Panhellenic Statistics Conference (Greek Statistical Institute), Spetses,Greece, 455–464

PilgrimDH and Cordery I (1993) Flood Runoff. In: Handbook of Hydrology, ed. D.R.Maidment, McGrawHill, IncSCS (Soil Conservation Service) (1972) Hydrology. SCS National Engineering Handbook, Section 4, U.S.

Department of Agricultural, Washington, DCSpecial Secretariat for Water (2012) The Implementation of Directive 2007/60/EC: Preliminary Flood Risk

Assessment. Greek Republic, Ministry of Environment, Energy and Climate Change, AthensSutcliffe JV (1978) Methods of flood estimation, a guide to flood studies report, report 49. Institute of Hydrology,

WallingfordTsakiris G (1995) Flood Runoff. In: Water Resources: I. Engineering Hydrology. ed. G. Tsakiris, Editions

Symmetria, Athens (in greek), pp 393–423Tsakiris G, Nalbantis I, Pistrika A (2009) Critical technical issues on the EU flood directive. Eur Water 25(26):

39–51U.S.B.R. (US Department of the Interior, Bureau of Reclamation) (1987) Design of small dams: a water

resources technical publication. United States Department of the Interior, Bureau of Reclamation,Washington, pp 25–37

Wetsphal DA (2001) Hydrology for drainage system design and analysis. In: Stormwater collection systemsdesign handbook. L. W. Mays (ed) McGraw-Hill

S206 S. Yannopoulos et al.

Yannopoulos S, Eleftheriadou E, Tzivani E, Giannopoulou Io (2012) Estimation of peak discharge in a smallungauged watershed based on IDF curves and synthetic unit hydrographs. Proceedings of the XIInternational Conference on Protection and Restoration of the Environment, eds. K.L. Katsifarakis, N.Theodossiou, C. Christodoulatos, A. Koutsospyros, Z. Mallios, July 3–5, Thessaloniki, Greece, 156–165

Yannopoulos S, Eleftheriadou E, Mpouri S, Giannopoulou Io (2014) Hydrologic and HydraulicProcedures for Engineering Application at Ungauged Watersheds: A Greek Case Study.Proceedings of the XII International Conference on Protection and Restoration of theEnvironment (eds. A. Liakopoulos, A. Kungolos, C. Christodoulatos, A. Koutsospyros), June 29to July 3, Skiathos Island, Greece, 43–49

Implementing the Requirements of the European Flood Directive S207